Be careful with your Sun disk RMAs!

We got a good reminder today that failed hard drives don’t go to the big drive graveyard in the sky when they are sent back on RMA. We got a replacement disk for a failed v20z drive in a cluster node and to our surprise when we popped it in instead of the box going into a PXE boot it started up RHEL AS 3 off the drive! Oh my! We took a quick look around and it’s obvious the disk came out of another v20z at Toshiba in Japan. There’s stuff on the drive. Logs. Code. Stuff I sure as hell wouldn’t want leaving my shop. We zero’ed the drive and image’ed it after our cursory glance.

Scary stuff.

So remember: Your dead disks are refurbished and sent back into the field. Obviously neither Sun or Seagate are erasing the platters. So don’t leave data on them! I’d love to hear how people ensure that any data left on the platters of an unresponsive drive are purged of data. How big of a magnet do you need to be sure the bits are sufficiently scrambled?

(Note: Someone will probably point out that this isn’t Sun’s fault, it’s Seagate’s. That’s true to a certain extent, but I will point out that I’ve *never* gotten a Seagate disk from NetApp that wasn’t zeroed.)


About this entry